Japan Eyes Cutting-Edge AI Under Preemptive Cyber Defense Policy

Japan is considering utilizing cutting-edge AI systems under its preemptive cyber defense policy, in the face of the risk of attacks by advanced models such as US startup Anthropic's Claude Mythos, a senior government official said.

National Cyber Director Yoichi Iida told Kyodo News in a recent interview that it is "inevitable" Japan will use sophisticated AI technology amid growing concerns over cyberattacks by AI models with high capabilities to detect security vulnerabilities. "It would be no surprise if the time between vulnerability discovery and exploitation were shortened to a fraction of a hundredth or a thousandth of what it used to be," he said.

The report specifically flags Claude Mythos, which is seen as highly capable of finding security vulnerabilities. After its April unveiling, Anthropic initially limited access to IT companies such as Google and certain financial institutions, over concerns the model could be misused for cyberattacks.

Iida's remarks follow Prime Minister Sanae Takaichi's Cabinet decision in March to enable the government, from October 1, to take "active cyber defense" measures that eliminate in advance the possibility of serious cyberattacks disrupting critical infrastructure, economic activity and people's livelihoods.

The government needs to consider "how to advance automation" in cyber defense operations while ensuring human involvement in the final decision-making process, Iida said, without clarifying possible concrete steps. He also said Japan should not rely on any one particular AI model, citing the US government's export control directive that suspended access to Mythos-class models in June — a suspension that already affected security vulnerability scanning operations for key government systems.

If cutting-edge AI models were exploited by nation-sponsored cyberattacks, such offensives could become "far more sophisticated," and their intensity is expected to increase, Iida added.

The key questions ahead: how Japan builds its own AI defense capability without depending on any single model, how AI automation is introduced once active cyber defense takes effect in October, and how human final decision-making coexists with high-speed automation.